numpy.random.shuffle

在做將caffe模型和預訓練的引數轉化為tensorflow的模型和預訓練的引數，以便微調，遇到如下函式：

1. def gen_data(source):

2. while True:

3. indices = range(len(source.images)) # indices = the number of images in the source data set

4. random.shuffle(indices)

5. for i in indices:

6. image = np.reshape(source.images[i], (28, 28, 1))

7. label = source.labels[i]

8. yield image, label

         之前卑鄙陋寡聞，不知道這個用法，按照字面上的意思是打亂，那麼這裡就應該是讓訓練資料集中的資料打亂順序，然後一個挨著一個地(for i in indices)生成訓練資料對。下面就從docs.scipy.org中查到的random.shuffle的用法：

numpy.random.shuffle(x)

Modify a sequence in-place by shuffling its contents.

舉例

python>>>

>>> arr = np.arange(10)
>>> np.random.shuffle(arr)
>>> arr
[1 7 5 2 9 4 3 6 0 8]

This function only shuffles the array along the first index of a multi-dimensional array（多維矩陣中，只對第一維（行）做打亂順序操作）:

python>>>

>>> arr = np.arange(9).reshape((3, 3))
>>> np.random.shuffle(arr)
>>> arr
array([[3, 4, 5],
       [6, 7, 8],
       [0, 1, 2]])This function only shuffles the array along the first index of a multi-dimensional array:

參考：·[1] https://docs.scipy.org/doc/numpy/reference/generated/numpy.random.shuffle.html#numpy-random-shuffle

            [2] https://github.com/ethereon/caffe-tensorflow/blob/master/examples/mnist/finetune_mnist.py

https://blog.csdn.net/jasonzzj/article/details/53932645?utm_source=copy